Alzheimer's disease (AD) is the sixth leading cause of death among people age 65 or older in the United States and the number of people with AD is projected to triple by 2050. Despite substantial growth in palliative and hospice care over the past decade, many people with AD receive suboptimal care at the end of life characterized by aggressive medical interventions and repeated hospitalizations. The proposed study will examine the relation of AD and mild cognitive impairment (MCI) to the use of health care services in the last three years of life. Specifically, we hypothesize that compared to people with no cognitive impairment (NCI) and MCI, people with AD will have fewer hospitalizations, emergency room visits, and home health care visits and less use of intensive medical procedures more than one year before death, but that this pattern will reverse in the last year of life. Also, people with AD will be less likely than those with NCI or MCI to use hospice services. We will investigate how specific factors of comorbid health conditions, physical function, nursing home residence, and race modify the relation of AD and MCI to the pattern of health services utilization in the last three years of life. Our general hypothesis is that the presence and nature of comorbid conditions (e.g., cancer versus end organ failure) and level of physical function will explain most of the relation of AD and MCI to health services utilization. Nursing home residence and race will influence the timing, intensity, and types of health services used over time, and the effect of nursing home residence, but not race, will differ across diagnostic groups. We will use rigorously collected longitudinal data from a general population sample of residents age 65 or older of an urban biracial community (CHAP) linked to Medicare claims data and mortality information to achieve the study aims with economy of resources. The sample will include 2154 persons, alive and dead, who have undergone clinical evaluation and received a diagnosis of AD (n=625), MCI (n=655), or NCI (n=874). The CHAP population has several characteristics that make it well-suited to address the specific aims of the proposed study. First, we will have uniformly ascertained diagnoses of AD and MCI in a well-characterized population that will permit us to quantify the relation of cognitive function to health services utilization in an unbiased manner. Second, we will have rigorously collected longitudinal data from a population-representative cohort with high rates of baseline (79%) and follow-up (90%) participation. This will permit us to not only compare health services utilization across diagnostic groups and between those who died and those who did not die, but also to make within-person comparisons over time. Third, the CHAP sample is 54% black and 46% white with a wide distribution of socio-economic status within each racial group so we can examine racial differences in health services utilization over time. We have achieved linkage for 90% of CHAP clinical evaluation participants and their Medicare enrollment and claims data for 1993 to 2002. As part of the proposed study, we will purchase additional Medicare enrollment and claims data for 2003 through 2007. PUBLIC HEALTH RELEVANCE: A better understanding of the types of health care services used by people with Alzheimer's disease at the end of life and the interaction of Alzhiemer's disease with other health problems could lead to both more appropriate use of limited health care resources and more appropriate end of life care for persons with Alzheimer's. Data from the proposed analysis can be used to guide clinical practice and health policy decisions about end of life care for people with Alzheimer's, potentially resulting in improved care for thousands of patients with Alzheimer's disease and their families and better public health resource utilization.